The goal of this project is to identify and evaluate factors that may influence variability in response in laboratory studies. For example, we found that differences in scoring criteria was the major factor that produced inter-scorer and inter-laboratory variability for the mouse epididymal sperm aneuploidy assay. This finding emphasizes the importance of strict technical controls for this assay, including controlling slide preparations for treatment-induced reductions in sperm count and coding slides to avoid subtle scorer bias. Another investigation demonstrated an association between lung fibrosis and inflammation and the occurrence of adrenal pheochromoctyoma in male Fisher 344 rats. This association (which was evident both in dosed and control animals) may help explain the variability in pheochromocytoma incidence observed from study to study and between dosed and control groups. We also evaluated the inter-laboratory variability in response in the uterotrophic assay for a high-potency reference agonist, ethinyl estradiol. Two model systems were compared: the immature female rat and the adult ovariectomized rat. Both models appeared to be robust, reproducible and transferable across laboratories for the single compound evaluated. Finally, we compared the performance of the Peto and Poly-k tests for detecting rodent carcinogenicity. If cause of death can be reliably determined, then the Peto test is reasonable. However, if such information is inaccurate, then the Peto test can be quite misleading. The Poly-k test does not require cause of death information, and is thus an attractive alternative to the Peto test when accurate cause of death information is unavailable.